LAMMPS Integration

[JaGeo]

Are there currently any plans on integrating LAMMPS?
I think there would also be many users. As there atomate workflows, I assume there might be plans.

[utf]

I'm not sure if anyone is working on LAMMPS at the moment, but this is definitely a feature I'm very interested in.

[JaGeo]

Maybe, we can talk about it next week and check if anyone has plans in this regard! My group would be interested as well.

[mjwen]

+1 for LAMMPS integration. It seems the LAMMPS workflow in atomate1 is dead because of a lack of maintenance?

[davidwaroquiers]

@gbrunin is currently working on a Lammps in pymatgen for the inputs. He has some developments (based on earlier work by @manaslkht and @rkingsbury) for which he will soon open an WIP PR in pymatgen (see https://github.com/gbrunin/pymatgen/tree/htmd/pymatgen).

[ml-evs]

After conversations with @gbrunin, @davidwaroquiers, @JaGeo, @gmrigna and @QuantumChemist (couldn't find your GitHub handle before!), I've started some work towards atomate2 LAMMPS integration over at Matgenix/atomate2 (see rough start in Matgenix/atomate2#1).

This may involve the parallel development of a new pymatgen LAMMPS add-on package (see Matgenix/pymatgen-io-lammps and discussion at materialsproject/pymatgen#2692 / materialsproject/pymatgen#2754). Once we have a v0.1 of that package, I will make a PR here with a MWE, if that is still desired!

[ml-evs]

This may involve the parallel development of a new pymatgen LAMMPS add-on package (see Matgenix/pymatgen-io-lammps and discussion at materialsproject/pymatgen#2692 / materialsproject/pymatgen#2754). Once we have a v0.1 of that package, I will make a PR here with a MWE, if that is still desired!

I spoke to soon, looks like the IO stuff will just be updated in pymatgen directly; this makes things slightly more cumbersome for development here, as the LAMMPS part will need an as of yet unreleased pymatgen version, so happy to keep it as a separate atomate add-on if that is desired...

[JaGeo]

@ml-evs As I was just seeing this: I don't see an issue with just asking for a release of pymatgen as soon as the lammps interface is merged. 🙂 This should be fast.

[ml-evs]

Another dramatic twist in the tale... I think we are now planning on adding LAMMPS workflows as an atomate2 add-on (I believe there were offline discussions on this between @utf and @davidwaroquiers). I've started development in another new repo, https://github.com/matgenix/atomate2-lammps and will raise a separate issue about add-on packages shortly.

[utf]

I'm going to close this since https://github.com/matgenix/atomate2-lammps is now a thing. It looks like there haven't been many updates though, so if someone wanted to take that on (and potentially incorporate it into atomate2 main, then that could be an option).

[JaGeo]

With the increasing importance of ML potentials, LAMMPS is probably more interesting than ever